sentiment analysis question in 2 parts
Hi everyone!
I am not a programmer so this will likely come across as a very naive set of two questions but here goes:
1. should I be using Rosette or Aylien for sentiment analysis of a Twitter feed?
2. i signed up for the Aylien trial period, but am several days into the 14-day free period and am beginning to panic since i cannot make heads or tails over how to make it work - as stated with the intention of connecting it up to a real time Twitter feed - then to obviously link it up to RapidMiner.
NOTE: I do not read python or any other command line language which is why i acquired RapidMiner because it allows me to work in a "no-code" environment.
Your help, your facile hand-holding, is VERY much appreciated! Richard
Find more posts tagged with
Sort by:
1 - 5 of
51
Both extensions are great for sentiment analysis. However, before you pay have a look at some much simpler and free extensions to do the same. One is a Dictionary-Based Sentiment operator from the Operator Toolbox extension. Another is a Wordnet extension which uses a free Wordnet dictionary. However, you can also do a fast non-dictionary based sentiment analysis by relying on the the algorithmic stemming and some word counting to come up with your own sentiment scores.
Very helpful. Looking to do a non-dictionary based sentiment analysis. So, you say I can rely on the algorithmic stemming and some word counting to come up with my own sentiment score? Questions 1, 2 & 3. first, where in Studio do I find this "algorithmic stemming," second, how do i apply it to do the word counting to come up with my own sentiment scores, lastly will this approach work with a real time Twitter feed and not be restricted only to static databases? Thanks again! Your help is most, most sincerely appreciated. Richard
I'd probably first try the existing sentiment analysis operators, such as those suggested above, especially the built in operator mentioned by @Telcontar120 (RM is changing so fast that I forgot that one). You can save yourself a lot of sweat in the process. The algorithmic solution also depends on a "dictionary" of positive and negative words but it uses standard stemmers, such as Snowball, Potter or Logins (all in Text Processing extension). Algorithmic stemming would be useful if your sentiment lists are huge, but it is not as accurate as those based on a proper dictionary. There is an example process I posted in one of the previous discussions, which @mschmitz calls old fashioned and obsolete, but which shows the ropes of sentiment analysis:
https://community.rapidminer.com/discussion/comment/64839#Comment_64839
https://community.rapidminer.com/discussion/comment/64839#Comment_64839
Good to hear that they're not offering any support because i certainly wasn't receiving any. Will switch to the Extract Sentiment Operator inside RapidMiner or alternatively Rosette or MonkyLearn if needed, as things progress from there. Short term goal right now is just to get a POC on an intuition.
Sort by:
1 - 5 of
51
Both extensions are great for sentiment analysis. However, before you pay have a look at some much simpler and free extensions to do the same. One is a Dictionary-Based Sentiment operator from the Operator Toolbox extension. Another is a Wordnet extension which uses a free Wordnet dictionary. However, you can also do a fast non-dictionary based sentiment analysis by relying on the the algorithmic stemming and some word counting to come up with your own sentiment scores.
Warning, the Aylien extension is no longer maintained by Aylien, so if you run into any problems you won't have any help solving them. I would recommend switching to Rosette or MonkeyLearn, or just using the native Extract Sentiment Operator inside RapidMiner.
Very helpful. Looking to do a non-dictionary based sentiment analysis. So, you say I can rely on the algorithmic stemming and some word counting to come up with my own sentiment score? Questions 1, 2 & 3. first, where in Studio do I find this "algorithmic stemming," second, how do i apply it to do the word counting to come up with my own sentiment scores, lastly will this approach work with a real time Twitter feed and not be restricted only to static databases? Thanks again! Your help is most, most sincerely appreciated. Richard
I'd probably first try the existing sentiment analysis operators, such as those suggested above, especially the built in operator mentioned by @Telcontar120 (RM is changing so fast that I forgot that one). You can save yourself a lot of sweat in the process. The algorithmic solution also depends on a "dictionary" of positive and negative words but it uses standard stemmers, such as Snowball, Potter or Logins (all in Text Processing extension). Algorithmic stemming would be useful if your sentiment lists are huge, but it is not as accurate as those based on a proper dictionary. There is an example process I posted in one of the previous discussions, which @mschmitz calls old fashioned and obsolete, but which shows the ropes of sentiment analysis:
https://community.rapidminer.com/discussion/comment/64839#Comment_64839
https://community.rapidminer.com/discussion/comment/64839#Comment_64839
Good to hear that they're not offering any support because i certainly wasn't receiving any. Will switch to the Extract Sentiment Operator inside RapidMiner or alternatively Rosette or MonkyLearn if needed, as things progress from there. Short term goal right now is just to get a POC on an intuition.